In the field of scanning electron microscope (SEM) and the related industrial fields which employ electron microscope principle to observe a sample, such as defect review and defect inspection for yield management in semiconductor manufacture, getting a specimen imaging with high resolution and low radiation damage has been required and pursued.
The only remedy to reduce radiation damage on specimen is to use a low-energy (or typically called as low-voltage in SEM field) electron beam scanning (<5 keV) which limits the beam penetration beneath the specimen surface and the residual charging on the specimen surface. However, the resolution will become worse because a low-energy electron beam forms a probe spot larger than a high-energy electron beam.
The probe spot diameter on specimen surface is determined by electron source size, spherical and chromatic aberrations, diffraction and Coulomb effect in the imaging system. For a low-energy electron beam, the achievable smallest probe spot size is limited by diffraction disc due to its larger Broglie wavelength λ and chromatic aberration due to its larger relative energy spread dV/V0. Both are respectively shown in equations (1.1) and (1.2). Here CCA is chromatic aberration coefficient, V0 and dV are electron energy and energy spread, and α is beam half angle. Obviously, to reduce probe spot size, reducing energy spread is another option as well as reducing chromatic aberration coefficient.
                    Dd        =                                            0.61              ·                              λ                α                                      ⁢                                                  ⁢            where            ⁢                                                  ⁢            λ                    =                      12.26                                          V                0                                                                        (        1.1        )                                Dc        =                              1            2                    ·                      C            CA                    ·          α          ·                      dV                          V              0                                                          (        1.2        )            
Energy spread of an electron beam comes from the original energy spread generated when electrons are emitted from an electron source and the imposed energy spread generated by statistic interaction among electrons on the way from the source to destination (Boersch effect). Electron energy distribution usually has a shape with a long tail, and the energy spread of a beam is typically expressed in FWHM (Full Width Half Maximum). For Schottky Field Emission Source which is widely used in Low-Voltage SEM (LVSEM), the energy spread dV at cathode is 0.3 eV, and at gun exit it is increased to 0.5-1 eV in dependence upon the beam current. For a low-energy electron beam such as 1 keV, this amount of the energy spread means a relative energy spread dV/V0 much larger than a high-energy beam such as 10 keV.
Many solutions have been provided to reduce the energy spread dV before electrons land on the specimen. In these solutions, magnetic and/or electrostatic deflectors (such as Alpha filter, omega filter and Wien filter) and electrostatic round lenses (such as U.S. Pat. No. 7,034,315) are taken as dispersion elements. These elements in common generate deflection dispersion when deflecting an electron beam. Among all these solutions, only Wien filter has a straight optical axis and does not deflect electrons with normal energy away from the optical axis. This characteristic makes Wien filter is easier in provision and generates no off-axial aberrations which are actually impossible to be completely compensated, and therefore many solutions are provided based on Wien filter.
In a fundamental configuration of a standard Wien filter as shown in FIG. 1, an electrostatic dipole field E in X direction and a magnetic dipole field B in Y direction are superposed perpendicularly to each other, and both are perpendicularly to a straight optical axis Z. An electron beam propagating along the optical axis Z goes through the Wien filter. Wien Condition is only true for the electrons moving in Z direction with a velocity ν0 as shown in equation (1.3), where the net Lorenz force F on each electron is zero. For the electron moving in Z direction with a velocity deviation δν from velocity ν0, it obtains a non-zero net Lorenz force in X direction as shown in equations (1.4) or (1.5) expressing in normal electron energy V0 and energy derivation δV, and will be deflected in X direction and therefore is diverted away from Z direction. Here e and m are respectively charge and mass of an electron. The deflection angle α depends on energy deviation δV and the deflection power K which is related to the magnetic field B and normal energy V0, as shown in equation (1.6). Hence, the Wien filter generates deflection dispersion, and the deflection power K represents the dispersion strength. For sake of clarity, the deflection power K and the deflection direction are respectively called as dispersion power and dispersion direction here. In this case no off-axial aberrations occur for the electrons with velocity ν0.
Wien Condition:F(ν0)=Fe+Fm=0 or E=ν0·B  (1.3)Dispersion:F(ν0+δν)=−e·δν·B  (1.4)
                              F          ⁡                      (                                          V                0                            +                              δ                ⁢                                                                  ⁢                V                                      )                          =                              -            e                    ·                                    δ              ⁢                                                          ⁢              V                                                      2                ·                m                ·                                  V                  0                                                              ·          B                                    (        1.5        )            α=K(B,V0)·δV  (1.6)
For each electron with normal energy but not moving in YOZ plane, it gains a potential change from the electrostatic field. Therefore, its velocity will be different from ν0 when it passes through the Wien filter as shown in equation (1.7) and will obtain a non-zero net Lorenz force as shown in equation (1.8). The net Lorenz force is proportional to electron position x, so a focusing effect in X direction (dispersion direction) exists. The focusing effect in dispersion direction will generate an astigmatic focusing, and simultaneously reduce deflection angles of the off-axis electrons. The latter implies a dispersion power reduction.
                                          υ            0                    -          υ                =                  -                                    e              ·              E              ·              x                                      m              ·                              υ                0                                                                        (        1.7        )                                          F          ⁡                      (            x            )                          =                              -                                                            ⅇ                  2                                ·                B                ·                E                                                              2                  ·                  m                  ·                                      V                    0                                                                                ·          x                                    (        1.8        )            
The Wien filter has been employed as a monochromator or an energy filter in many ways, wherein energy depending filtering and energy-angle depending filtering are two typical ways. In energy depending filtering shown in FIG. 2a, a beam 2 from an electron source 1 is focused by a round lens 10 and/or Wien filter 11 itself (such as U.S. Pat. Nos. 6,452,169, 6,580,073, 6,960,763 and 7,507,956), and forms an astigmatic image on an energy-limit aperture 12. The electrons with energy V0 forms a sub-beam 3 which is focused onto optical axis, while the electrons with energy V0±δV respectively form sub-beams 4 and 5 which will be respectively deflected in ±X direction and focused away from the optical axis. As a result, inside the beam 2, all the electrons whose energy deviations are within ±δV will pass through the aperture 12 and the rest will be blocked out.
As a significant advantage, the energy depending filtering will cut off the long tail of electron energy distribution completely. The long tail of energy distribution generates a background in the image and deteriorates the image contrast. As an unignorable disadvantage, the energy depending filtering increases the source size. The image of the electron source 1 on the aperture 12 is the source for the following electron optics, whose size is actually determined by the aperture size. However the practicable aperture size at present (≧100 nm) is much larger than the size of the original source 1 (virtual source of Schottky Field Emission Source is about 20 nm). In addition, the image on the aperture 12 is a crossover of all electrons, which enhances electron interaction that generates additional energy spread. Although, an astigmatic image is better than a stigmatic image in terms of the electron interaction.
In energy-angle depending filtering (for example, U.S. Pat. Nos. 6,489,621, 7,679,054 and 5,838,004) as shown in FIG. 2b, a beam 2 from an electron source 1 passes through the Wien filter 11. The electrons with energy V0 forms a sub-beam 3 which goes straight, and electrons with energy V0±δV respectively form sub-beams 4 and 5 which will be respectively deflected in ±X direction. The position of each electron on energy-angle-limit aperture 12 depends on its energy and incident angle into the Wien filter 11 as well. Therefore, the aperture 12 not only blocks out all the electrons whose energy deviations are not within ±δV, but also some of the electrons which have larger incident angle even so their energy deviations are within ±δV.
The deflection angle α with respect to the energy deviation δV must be at least larger than double incident half angle β to clear filter out charged particles with energy deviation δV. This requires dispersion power of the Wien filter to be strong enough or divergence of the incident beam to be small enough. Increasing dispersion power of the Wien filter will increase deflection angle, but at the same time enhance its focusing effect which will in turn decrease the deflection angle and limits its achievable maximum of deflection angle. Restraining divergence of the incident beam will either limit the beam current or enhance electron interaction which in turn increases energy spread of the beam. Another unignorable disadvantage is the original source 1 is changed to be a larger virtual source from 14 to 15 for the following electron optics.
Many improvement methods have been provided to solve the problems mentioned above. For the energy-angle depending filtering, one method is to use a round lens to image the original source onto the Wien filter center (such as U.S. Pat. No. 7,468,517). This minimizes the Wien filter effect on source size, but adds a real crossover. Another method is to use a second Wien filter to compensate the residual effect of the first Wien filter (such as U.S. Pat. No. 6,489,621, 7,679,054). Although this method does not generate a real crossover, it generates a virtual crossover far away from the following electron optics which will incur large aberrations due to a large increase in beam size.
For the energy depending filtering, the methods with one (as shown in FIG. 3a and FIG. 3b) or more additional Wien filter 21 to compensate the residual effect of the first Wien filter 11 after the energy-limit aperture filter 12 are provided in many documents (such as U.S. Pat. Nos. 6,960,763, 6,580,073 and 7,507,956). In these solutions, a stigmatic and dispersion-free crossover 7, i.e. an additional real crossover after the first real astigmatic crossover 6 at the energy-limit aperture 12, is formed after the last Wien filter 21. This not only increases the energy spread after the energy filtering in terms of electron interaction, but also lengthens the total length of SEM by at least the length 8 of the monochromator.
The present invention will provide a solution to solve the problems in the energy depending filtering and energy-angle depending filtering. Instead of forming a real stigmatic crossover of an incident charged particle beam after the monochromator, it forms a virtual stigmatic and dispersion-free crossover inside the monochromator. Thereafter it provides an effective way to improve the imaging resolution of low-Voltage SEM and the related apparatuses which are based on LVSEM principle.